Most on-page checklists treat the meta title and the H1 as the same field with two names. On the Unscripted SEO podcast, Bradley Benner of Semantic Mastery and Semantic Links argued they are two different jobs for two different readers — and that collapsing them into one string is why so many pages are technically optimised and commercially dead.
The Two Audiences On Every Page
Every page you publish is read twice. Once by a parser that is trying to work out what entities the page is about and how they relate to each other, and once by a human who has just clicked through and is deciding in about two seconds whether to stay.
Benner’s framing is blunt about which field serves which reader:
“the meta title is for the bots.”
Bradley Benner, Semantic Mastery / Semantic Links
“But the H1 is for the human visitor.”
Bradley Benner
That is the whole split. Once you accept it, the rewrite work becomes mechanical rather than creative.
What The Meta Title Has To Do For A Model
Benner teaches the SEO title as an entity association statement — a semantic triple. Subject, predicate, object. The brand or page entity, the relationship, the thing it relates to.
“For search, you should optimize the SEO title in a very strict semantic triple format.”
Bradley Benner
In practice that means the meta title carries the named entity, the qualifier that disambiguates it, and the query-shaped phrase a model can match against. It is not where you put your cleverness. It is where you put your nouns.
If you are already working through the quality rater guidelines tactic, this slots in directly ahead of it: get the entity statement right first, then judge the page against rater criteria.
What The H1 Has To Do For A Human
The H1 is not a second chance to repeat the keyword. Benner describes it as a restatement — same entities, different register.
“But then the H1 becomes a restatement of the entities that were referenced in the SEO title.”
Bradley Benner
“That should be compelling and a lot more conversational, but still referencing the entities that were mentioned in the SEO title, but using variants, right?”
Bradley Benner
Variants matter. Repeating the meta title verbatim in the H1 wastes the one piece of above-the-fold copy that is allowed to sound like a person wrote it, and it gives the parser no additional surface forms to associate with the entity.
A Fifty-Page Rewrite Pass You Can Run This Week
This is a spreadsheet job, not a project. Export your top fifty URLs by impressions. Four columns: URL, current title, new machine-facing title, new human-facing H1.
- Name the entity. What is this page actually about — a service, a product, a location, a person? Write it as a noun phrase.
- Write the triple. Entity, relationship, object. Strict. Boring is fine.
- Write the H1 second. Same entities, conversational phrasing, variant wording.
- Check for collisions. If two pages produce the same triple, you have a consolidation problem, not a title problem.
- Ship in batches of ten so you can attribute movement.
The content brief SOP is the natural place to bake this in permanently — if the brief demands both fields, writers stop shipping one string twice.
How To Tell If The Split Worked
Two signals, watched separately. The meta title change should move impressions and the range of queries a URL surfaces for. The H1 change should move engaged sessions and scroll depth, because it is doing conversion work, not retrieval work.
If impressions move and engagement does not, your triple is good and your H1 is still robotic. If engagement moves and impressions do not, the reverse. Watching one blended metric hides both.
Benner’s wider position on measurement is worth holding alongside this — he is unsentimental about third-party scores generally:
“And so I totally believe 100% wholeheartedly that third-party metrics are useless.”
Bradley Benner
For the underlying crawl and rendering questions this rewrite assumes are already solved, see the technical SEO and site fundamentals guide.